Text::RecordParser - read record-oriented files

      use Text::RecordParser;
      my $p = Text::RecordParser->new;

      # Split records on two newlines

      # Split fields on tabs

      # Skip lines beginning with hashes
      $p->comment( qr/^#/ );

      # Trim whitespace

      # Use the fields in the first line as column names

      # Get a list of the header fields (in order)
      my @columns = $p->field_list;

      # Extract a particular field from the next row
      my ( $name, $age ) = $p->extract( qw[name age] );

      # Return all the fields from the next row
      my @fields = $p->fetchrow_array;

      # Return all the fields from the next row as a hashref
      my $record = $p->fetchrow_hashref;
      print $record->{'name'};

      # Get all data as arrayref of arrayrefs
      my $data = $p->fetchall_arrayref;

      # Get all data as arrayref of hashrefs
      my $data = $p->fetchall_arrayref( { Columns => {} } );

      # Get all data as hashref of hashrefs
      my $data = $p->fetchall_hashref('name');

    This module is for reading record-oriented data. The most common example
    have records separated by newlines and fields separated by commas or
    tabs, but this module aims to provide a consistent interface for
    handling sequential records in a file however they may be delimited.
    Typically this data lists the fields in the first line of the file, in
    which case you should call "bind_header" to bind the field name. If the
    first line contains data, you can still bind your own field names via
    "bind_fields". Either way, you can then use many methods to get at the
    data as arrays or hashes.

    This is the constructor. It takes a hash of optional arguments. Each
    argument can also be set through the method of the same name.

    * filename
        The path to the file being read. If the filename is passed and the
        fh is not, then it will open a filehandle on that file and sets "fh"

    * comment
        A compiled regular expression identifying comment lines that should
        be skipped.

    * data
        The data to read.

    * fh
        The filehandle of the file to read.

    * field_separator
        The field separator (default is comma).

    * record_separator
        The record separator (default is newline).

    * field_filter
        A callback applied to all the fields as they are read.

    * header_filter
        A callback applied to the column names.

    * trim
        Boolean to enable trimming of leading and trailing whitespace from
        fields (useful if splitting on whitespace only).

    See methods for each argument name for more information.

    Alternately, if you supply a single argument to "new", it will be
    treated as the "filename" argument.

    Takes an array of field names and memorizes the field positions for
    later use. If the input file has no header line but you still wish to
    retrieve the fields by name (or even if you want to call "bind_header"
    and then give your own field names), simply pass in the an array of
    field names you wish to use.

      $p->bind_fields( qw[ name rank serial_number ] );

    Takes the fields from the next row under the cursor and assigns the
    field names to the values. Usually you would call this immediately after
    opening the file in order to bind the field names in the first row.

      my $name = $p->extract('name');

    Takes a regex to apply to a record to see if it looks like a comment to

      $p->comment( qr/^#/ );  # Perl-style comments
      $p->comment( qr/^--/ ); # SQL-style comments

    Allows a scalar, scalar reference, glob, array, or array reference as
    the thing to read instead of a file handle.

      $p->data( $string );
      $p->data( \$string );
      $p->data( @lines );
      $p->data( [ $line1, $line2, $line3] );
      $p->data( IO::File->new('<data') );

    It's not advised to pass a filehandle to "data" as it will read the
    entire contents of the file rather than one line at a time if you set it
    via "fh".

    Extracts a list of fields out of the last row read. The field names must
    correspond to the field names bound either via "bind_fields" or

      my ( $foo, $bar, $baz ) = $p->extract( qw[ foo bar baz ] );

    Reads a row from the file and returns an array or array reference of the

      my @values = $p->fetchrow_array;

    Reads a line of the file and returns it as a hash reference. The keys of
    the hashref are the field names bound via "bind_fields" or

      my $record = $p->fetchrow_hashref;
      print "Name = ", $record->{'name'}, "\n";

    Like DBI's fetchall_arrayref, returns an arrayref of arrayrefs. Also
    accepts optional "{ Columns => {} }" argument to return an arrayref of

      my $records = $p->fetchall_arrayref;
      for my $record ( @$records ) {
          print "Name = ", $record->[0], "\n";

      my $records = $p->fetchall_arrayref( { Columns => {} } );
      for my $record ( @$records ) {
          print "Name = ", $record->{'name'}, "\n";

    Like DBI's fetchall_hashref, this returns a hash reference of hash
    references. The keys of the top-level hashref are the field values of
    the field argument you supply. The field name you supply can be a field
    created by a "field_compute".

      my $records = $p->fetchall_hashref('id');
      for my $id ( keys %$records ) {
          my $record = $records->{ $id };
          print "Name = ", $record->{'name'}, "\n";

    Gets or sets the filehandle of the file being read.

      open my $fh, "<./data.csv";
      $p->fh( $fh );

    A callback applied to the fields identified by position (or field name
    if "bind_fields" or "bind_header" was called).

    The callback will be passed two arguments:

    1   The current field

    2   A reference to all the other fields, either as an array or hash
        reference, depending on the method which you called.

    If data looks like this:

      parent    children
      Mike      Greg,Peter,Bobby
      Carol     Marcia,Jane,Cindy

    You could split the "children" field into an array reference with the
    values like so:

      $p->field_compute( 'children', sub { [ split /,/, shift() ] } );

    The field position or name doesn't actually have to exist, which means
    you could create new, computed fields on-the-fly. E.g., if you data
    looks like this:


    You could write a field_compute like this:

        $p->field_compute( 3,
            sub {
                my ( $cur, $others ) = @_;
                my $sum;
                $sum += $_ for @$others;
                return $sum;

    Field "3" will be created as the sum of the other fields. This allows
    you to further write:

        my $data = $p->fetchall_arrayref;
        for my $rec ( @$data ) {
            print "$rec->[0] + $rec->[1] + $rec->[2] = $rec->[3]\n";


        1 + 3 + 5 = 9
        32 + 4 + 1 = 37
        9 + 5 + 4 = 18

    A callback which is applied to each field. The callback will be passed
    the current value of the field. Whatever is passed back will become the
    new value of the field. Here's an example that capitalizes field values:

      $p->field_filter( sub { $_ = shift; uc(lc($_)) } );

    Returns the fields bound via "bind_fields" (or "bind_header").

      $p->bind_fields( qw[ foo bar baz ] );
      my @fields = $p->field_list;
      print join(', ', @fields); # prints "foo, bar, baz"

    Returns a hash of the fields and their positions bound via "bind_fields"
    (or "bind_header").

    Gets and sets the token to use as the field delimiter. The default is a
    comma. Regular expressions can be specified using qr//.

      $p->field_separator("\t");     # splits fields on tabs
      $p->field_separator('::');     # splits fields on double colons
      $p->field_separator(qr/\s+/);  # splits fields on whitespace
      my $sep = $p->field_separator; # returns the current separator

    Gets or sets the complete path to the file to be read. If a file is
    already opened, then the handle on it will be closed and a new one
    opened on the new file.


    A callback applied to column header names. The callback will be passed
    the current value of the header. Whatever is returned will become the
    new value of the header. Here's an example that collapses spaces into a
    single underscore and lowercases the letters:

      $p->header_filter( sub { $_ = shift; s/\s+/_/g; lc $_ } );

    Gets and sets the token to use as the record separator. The default is a
    newline ("\n").

    To read a file that looks like this:


    Set the record and field separators like so:


    Remove leading and trailing whitespace from fields.

      my $trim_value = $p->trim(1);

    Ken Y. Clark <>

    Thanks to the following:

    * Benjamin Tilly
        For Text::xSV, the inspirado for this module

    * Tim Bunce et al.
        For DBI, from which many of the methods were shamelessly stolen

    * Tom Aldcroft
        For contributing code to make it easy to parse whitespace-delimited

    Copyright (c) 2003-4 Ken Y. Clark

    This library is free software; you can redistribute it and/or modify it
    under the same terms as Perl itself.

    Please use for reporting bugs.